Reduce engine size.

[minh-thanh]

Hi ALL !
Reduce engine size
I am designing an engine
Reducing the engine size, do you keep the ratio of all parts: gears, cylinder diameter, stroke, ... all structural details.? Or are you just trying to keep the shape of the engine the same as the large size engine?
Thanks.

 

[Ken I]

Generally you keep everything the same.

However you should keep in mind that scaling up or down introduces problems.

Areas increase/decrease to the square of the scale, volumes and mass to the cube of the scale and polar and mass moments to the 4th power - this last item is the most trouble.

Example:- If you take a 2 litre motor and double its dimensions - the volume / mass is 2x2x2 = 8 times bigger - so if the original engine was 2 litre and weighed 200kgs - your scaled up engine to twice as big (dimensionally) will become 16 litres and weigh 1.6 tonnes !

By the same token if you scaled it down 2:1 (half size) it would become 250cc and weigh 25kg.

In the case of the 2 litre - if it made 200 Horsepower (a hundred horsepower per litre - high performance) - scaling is going to have a strange effect - the double size engine is going to have the same mean effective pressure applied to 4 times the area and twice the stroke so the torque is going to be 8 times greater - however the mean piston speed cannot be increased (limits of lubrication capability were already at maximum on the 2 litre engine) so if our 2 litre motor was capable of 6000rpm our 16 litre motor will only be capable of 3000rpm so overall our power only increases 4 times to 800 horsepower not 1600 as you might have expected. So it achieves only half the specific horsepower - 50 horsepower per litre.

By the same token our half size 250cc model can do 12000 rpm and generate 50 horsepower or 200 horsepower per litre - that is why high performance engines have more smaller cylinders.

O.K. I'm talking theoretically here - in practice I don't think the 250cc would be that good or the 16 litre that bad - but always have these ratios in your mind. (Because of aspiration and carburation issues - atoms don't scale ! Flame speed and flame propagation in the engine remain the same regardless of its size etc. etc.)

Now we come to the tricky bit - the polar moment - the ability of a shaft to resist torque - is to the 4th power - so in the case of our 2 litre scaled up to 16 litre the torque increased by 8 times but the ability of the crank to resist the torque went up 2x2x2x2 = 16 times - so the crank becomes over-designed for the application and the journals could be reduced.

The opposite is going to happen with our 250cc scaled down engine - the torque is going to be 1/8th but the ability of the rotating parts to handle it is going to be 1/16th and therefore be much more highly stressed and is bound to fail.

You can see this on large cranks from marine engines - they look relatively "skinny" when compared to our normal frame of reference - a crankshaft from a car engine.
If you radically scale down such an engine - say 1/10th scale you are going to end up with a crank that is effectively only 10% of the original torsional design strength relative to its new size.

We generally don't want to actually derive extreme performance from a model - so you can get away with it - but be careful.

Assuming you want to make allowance - lets say the 2 litre's main crank bearings were Ø60mm and the big end journals Ø40mm then our scale sizes would be Ø30mm and Ø20mm which is actually too small for the design.
The true (compensated) scale for the Ø60 should be the 4th root of (60^4)/8 which scales down to Ø35.67 (not Ø30 as you might presume).and the Ø40 scales down to Ø23.8 (not Ø20 as you might presume). So we would round up to Ø36 & Ø24

(A simple way to calculate the 4th root is to take the square root twice.)

So torsionally stressed parts need to be slightly larger than scale on scaled down motors.

What you can also see is that a small change in diameter makes a big difference to polar moment - so if you scale down a shaft to say Ø6.9 then round it up to Ø8 - or apply the calculation - never round down. (The Ø8 shaft would be 80% torsionally stronger than Ø6.9).

You don't have to slavishly follow such "Strength Of Materials" type calculations but you should always have these rules at the back of your mind. Also bear in mind the actual strength of the materials you have chosen etc. etc.

In most cases, scaling down works to your favour in terms of strength in everything except torsionally loaded parts.

The inertia of a flywheel is also to the fourth power - so scaled down flywheels have considerably less inertia relative to the scaled down engine - err upwards on diameter and thickness when scaling down flywheels.

A final comment - atoms don't scale - so things like lubrication clearances remain the same and effectively scale up leakage, by-pass etc. in a model that has been scaled down. Hence frequent problems with compression and carburation etc. on small scale motors.

Regards, Ken

 

[tornitore45]

Excellent post Ken.

 

[minh-thanh]

Ken !
Thank you for your time and explanation .

 

[Ken I]

You're welcome - I edited in an additional comment on flywheels - also a fourth power problem.

Regards, Ken

 

[KenC]

And don't forget that if you half all dimensions of an i/c engine the displacement is 1/8 of the original, yet half the choke diameter of the carburettor inlet only reduces the air intake by 1/4 and not by 1/8, so halving all dimensions is not in every case workable. The carb needs to have 1/8 of the area of the original design.
Ken

 

[Steamchick]

An excelent explanation. The same principles apply to stiffness of lathe beds (and other machines) subject to the torsion of cutting. Another thread discussing parting-off expresses how some lathes are OK yet others are not. I mooted that the stiffness of the lathe bed is a significant factor for heavy cutting on smaller "hobby" lathes, where "industrial" lathes have no problem. The simple solution is to let the designers make all the calculations - which is what they are paid to do - then don't expect "amateur guesswork" to "know better". As an ex design engineer, I am always dismayed when excellent models don't work because the modelers don't do the Engineering. Calculations are absolute decisions, ignore them at your peril.
Bleat ended.
Enjoy your modelling.

 

[propclock]

Thank you for your time in posting this . I wasn't aware of the polar moment factor.
Good morning thoughts to digest today.

 

[peterl95124]

Ken I,
many thanks for the explanations, I had only thought as far as cylinder head bolt thickness and con-rod bolt thickness, which I think can be left scale (but don't scrimp, use grade-8 bolts), but not about torsional strengths and moments. Another item that doesn't scale is spark ignition. Fuel/air mixture requires a minimum spark energy to ignite, you can't scale down the voltage (to avoid spark-over in the distributor), and decrease plug-gap to make up for it, and still get ignition.
Peter L.

 

[Steamchick]

Just for an odd thought... Instead of scaling-down linear dimensions and getting it wrong, the "ideal" way is to design an engine of the required size and configuration (appearance, etc.) from scratch. The calculations are standards, that always work, and give the correct sizes for whatever size is selected. If you buy proprietary finished engines - like small aero-engines, etc. - then you will get parts that are designed for that SIZE of engine. E.G. a linearly scaled-down piston isn't right, but a proprietary Aero-engine piston of that same size has been designed to be right - at that size. And while it is good to make it yourself, surely the purist would re-design the parts to be Engineered to be correct at the scaled-down size?
As an example: I have a text book on Locomotive design - which has the calculations I use for working-out balancing of flywheels when I want to balance my model engines. - But maybe I am the "anorak that should be locked in the cupboard"? (I look at most models in "Engineering" shows and can't find the Engineering, just pretty scaled-down models).
Consider as another example, Martin Evan's designs of boilers: The external dimensions of some boilers may be based on "full-sized" boilers, but everything else is Engineered. - Material wall thicknesses, tube numbers, sizes, numbers and spacing of stays, etc. etc.
So maybe this is the example that one should aim to follow when doing their own "design" of scaled-down model?
But however you do it, ENJOY IT. - Best regards, K.

 

[peterl95124]

Steamchick, I like your approach also. I've been collecting "rules of thumb" for IC engine design (many came from Liston's "Aircraft Engine Design", McGraw-Hill, 1942) for a long time. For example piston-cylinder clearance should be about 0.001" per inch of bore, except 0.004" per inch of bore at the top above the first ring. Ring gap about 0.003" per inch of bore. And Journal-Bearing clearance. And I also have cam timing for a large number of designs in StrictlyIC plus some full size auto designs (throw out the widest and narrowest and average all the rest and you've got a good "general purpose" cam timing). And etc, etc, etc. Someday I'd like to publish it all in maybe Mike Rehmus' Model Engine Builder so there would be a permanent reference.

 

[Steamchick]

Good idea..... I guess?
But: Just to throw some spanners around the works.... as, while I am not an expert, I have with with some, and picked up a few ideas of what to do in life.
First, on valve tming. This is highly complex, but crude ideas will work, if to achieve any real "performance" level. Simply, the intake and exhaust gases are defined by a lot of parameters, and are not "simple gases". So consider the simple things first: a car engine may have a 4" bore, the model 3/8". Or 10 cm and 1cm... so is about 1/10th scale. Stroke also 1/10th scale. So displacement of model is 1/1000th of full size. Now you want the same piston speed, so you'll be running at 10 times the rpm of the full sized engine. Which in turn means 10 times the 1000th displacement = 1/100th of the average mass air flow drawn into the engine. Correct me if I am wrong as I am ad-libbing a bit here....! So if you want the same intake velocity of the intake gas (limited by the speed of sound at the pressure in the intake manifold!), then you'll want 1/100th of the cross-sectional area of manifold. Or 1/10th the diameter....MAGIC, as the linear scaling is good. Similarly, for the shock-waves in the intake and exhaust columns of gas, as at the same mean gas velocity (speed of sound) the 10 x frequency of the model will suit 1/10th of the pipe length....
But this is actualy more complex.... as gas and shock wave speeds are affected by pressure, and temperature, and both of those will be different on the model as at 1/10th scale you'll only have 1/100th of the heat generated, and (due to surface area scaling), 10 times the surface area per unit volume. (A 10 cm cube has 1000ccs, 600sq. cm. CSA, a 1cm cube has 1cc voltage and 6sq.cm. CSA.). This means 10 times more heat is transfered to cylinder and passage walls. Which changes the pressures and volumes of gas at all stages in the engine.
The valve timing has been tuned on the full-sized engine to otitis the performance for its purpose. But maybe your model is for idling on the bench? Or powering an aircraft? So the duty cycle for the model may be the same or similar to the real version. If the same, a similar valve timing may be a good starting point, but if really different from the original, a different valve timing will be much better. I.E. more suitable. So I am concerned that due to the complexity of the problems, your optimisation won't be very successful. It may be just something that works, by chance, rather than "by design".
But good luck, and hope that better engine designers can join the discussion and teach us all a better way.
I do like your plans and hope you succeed.
K

 

[peterl95124]

Steamchick, good thoughts about valves and porting. Some empirical evidence, the classic Volkswagon opposed-4 engine has two valves per cylinder and gets some HP, the Subaru opposed-4 engine with the same displacement but four valves per cylinder gets about twice the HP. IE scaling laws won't help you design a good high performance small engine, just plain facts. As far as cam design goes, I have a diagram that shows how the Toyota Prius Variable-Valve-Timing is adjusted for various driving conditions, the intake opening is delayed considerably for starting/stopping and idling, but is adjusted minutely around TDC for all other conditions. Cam timing is only "complicated" because we run full size engines at such various RPMs and loads, but with a fixed cam it can only be optimal for a single RPM and load. Since we don't measure the power output of model engines the cam design doesn't really make any difference except for two things, 1) no overlap of intake and exhaust at TDC is easier starting, 2) the timing does affect the sound of the engine although I don't have any hard data on that yet only anecdotal second hand evidence. I'm sure hoping the cam I'm grinding for my V-12 makes a good sound, its designed for zero intake-exhaust overlap because of a large volume induction system and I don't want any backfires !!! YMMV, Pete.

 

[Steamchick]

Thanks Pete, Interesting!
But I suggest that when comparing engines, you need to minimise variables. e.g. the difference between the VW and Subaru is a lot of other factors I think (Not knowing those specific details): Possible Cam/valve-timing, Fuelling system (manifolds and carb sizes), compression ratio, Displacement, and Valve sizes, as well as the 2-valve : 4-valve difference. I did some "Home experimenting" with Motor-cycle engines, as I had 500cc Triumph engines, carbs, cams, and exhausts to "ring the changes". e.g. When commuting I wanted the "trials engine" cam and timing and a single carb was OK for this. Just easy to ride with "apparently" good low-down torque for ease of moving away from the following traffic at Traffic Lights. And that was true with low and high compression pistons, with appropriate ignition timing and carburation. But the top-end performance was really lacking.... Which was improved by the "standard" cam (valve timing with increased overlap). But of course, for "thrashing around the roads", the best was a different cylinder head with bigger ports, sporty cams, higher compression of the limited edition "500C" top-end. (This was sold as a racing kit in the mid-1950s, and when I used it on a "standard" crankcase it ripped the crank-case apart! - It needed the later crankcase which had thicker walls and bosses all over). But I did learn "hands-on" that without variable cam timing, the factory produce a very good compromise of cam timing for commuting and highway use. My point for a book of "small" engines for models was that if you need the engine to replicate a decent performance for powering an aircraft, boat, car, etc. then you should consider a different cam and timing to an "idling" engine.
When I worked for a car manufacturer, the engine test for valve test was a long period of idling. Actually, although most testing was planned for the Engine Dynos, it was the one test that could be done on a simple test rig - just like a rig for demonstrating an engine at Model shows. I.E. a sturdy frame, fuel tank, ignition and local cooling system. The only loads on the engine were the Alternator for Ignition, battery charging and electric fan for the radiator. The reason for Idle testing for cam and valve gear wear, is that at high speeds, the dynamic tuning of the valve system causes cam followers to slide-up then "leap over" the crests of cams reducing contact pressure considerably on the lowering side, before catching the full pressure when the cam is nearing the bottom displacement region. So at speed, the high points for wear are when the valve springs are at their lowest pressure for cam to operate, whereas at idle, the cams wear on the crests, due to the higher spring pressure on the oil-film and cam face.
Complex stuff this "Engineering", I reckon!
Have you read "Tuning for Speed" by Phil Irving, or any other books of engine tuning? - there must be thousands out there. The same fundamental rules apply to small engines, and "idling" engines, so there is no need to re-invent the wheel, just maybe explain different applications and designs?
Enjoy,
K

 

[Jasonb]

A lot will depend on what you want out of the scaled engine, if it is performance then there are a few more things to be considered as mentioned above, if just running for enjoyment and display then as per the original post just reduce everything but the same factor, also works when you want to enlarge an engine too. I've done 15 or 16 engines now that are a different size than the original, all run OK and all I did was reduce or increase by a set amount eg let 1mm = 1/16" to get a smaller engine or let 1mm = 1/32" if you want larger.

 

[minh-thanh]

What you can also see is that a small change in diameter makes a big difference to polar moment - so if you scale down a shaft to say Ø6.9 then round it up to Ø8 - or apply the calculation - never round down. (The Ø8 shaft would be 80% torsionally stronger than Ø6.9).

Regards, Ken

The Ø8 shaft would be 80% torsionally stronger than Ø6.9
Hi Ken I !
Do you (or someone else) have a calculation formula for this??
I want to know more about it
Thank you !

 

[Ken I]

The Ø8 shaft would be 80% torsionally stronger than Ø6.9
Hi Ken I !
Do you (or someone else) have a calculation formula for this??
I want to know more about it
Thank you !

See :-

materials - torsionload.

Standard strength of materials calculations - for a shaft in torsion its to the 4th power. Bending 3rd power, tension or compression 2nd power.

So without doing any calculation of actual stress then an 8mm vs a 6.9 is (8^4) / (6.9^4) = 1.8070 or 80.7% stronger.

Just being aware of the ratios is all you mostly need.

Regards, Ken I

 

[Steamchick]

Thanks Ken 1 - you beat me to it. (And you are a Proper Engineer that explains things so I can understand them.).
Minh Thanh, does this help you?
K2

 

[minh-thanh]

Ken I
Thank you .

Minh Thanh, does this help you?
K2

I'm trying to understand
If you have more information...please

 

[Steamchick]

https://www.engineeringtoolbox.com/torsion-shafts-d_947.html
Does this translate? I appreciate that you are not a native English speaker, but perhaps you also understand French, if you can't translate into Vietnamese? Yours is a country I know very little about, but was it once called French Indo-China? - I may be wrong, so please teach me.

Anyhow:
The shear stress in a solid circular shaft in a given position can be expressed as:

τ = T r / J (1)

where

τ = shear stress (Pa, lbf/ft2 (psf))

T = twisting moment (Nm, lbf ft)

r = distance from centre to stressed surface in the given position (m, ft) (e.g. at the surface, this is half the diameter of the material when in pure torsion, but less as you move towards the centre of rotation = axis of torque).

J = Polar Moment of Inertia of Area (m4, ft4)

AND J = π (r 4th ) / 2: Which leads to Ken 1 s comment "torsional stress is to the 4th power" (Polar Moment of area).

Of course, you can be sure it is more complex as soon as you apply bending stresses, and stress concentration factors, etc. But even considering these things in isolation it is fairly easy to appreciate the simple ratios of each individual effect, and improve a design accordingly. (Or take an Engineering degree, and learn properly, which I missed somehow? - C'est la vie!). I find the Engineering toolbox explains what I think I know, better than I can. The Engineering ToolBox
Regards,
Ken